Seedance 1.0: Exploring the Boundaries of Video Generation Models (ByteDance Seed technical report)
2025年6月10日 · 自动从公开来源采集，未经人工审核。
这篇内容目前只有 英语 原文版本。
What happened
ByteDance Seed published the Seedance 1.0 technical report on 10 June 2025 (revised 28 June 2025), describing a video foundation generation model that generates a 5-second video at 1080p resolution in 41.4 seconds on an NVIDIA L20, with about a 10x inference speedup from multi-stage distillation and system-level optimisations. The report lists 44 authors, led by Yu Gao.
Why it may matter to you
Seedance 1.0 is ByteDance's video generation model; the report is the company's own disclosure of what it built and how it performs, which is the starting point for any question about what data it was trained on.
What is not yet known
The report's abstract does not name the training data. Whether the model was trained on creators' work, and on what terms, is not stated here.
来源
- Seedance 1.0: Exploring the Boundaries of Video Generation Models — arXiv (primary)